Search That Matches the Customer, Not Just the Query

On-site search that ranks, recommends and sells, in their language.

Trusted by

Matching Finds the Query. Relevance Finds the Customer

Matching a query to a catalogue is a solved problem. Algolia, Elasticsearch and others do it well, and we build on them too. But matching isn't relevance. The same query means different things to different customers; the result that's right for one is wrong for another.

The real value of search shows when results are ranked for the person, what they like, what they've done, what fits them now. That's when relevance climbs, the right products surface firs

We Build Search That Sells, Not Just Finds

Relevant results for each customer, ranked on intent, behaviour and language, with recommendations woven in

Vernacular & Voice Search

Search that understands Hinglish, code-mixing, voice and regional languages, so customers find what they want faster.

Intent & Behaviour Ranking

Intent-led ranking that blends search, behaviour and recommendations to surface the right products first.

Merchandised Recommendations

Customers find more than they searched for — your recommendation logic, rendered in the results, with rules you control.

The world's largest sporting goods retailer

Built and tuned Algolia search with vernacular synonyms and typo-tolerance, matched to how Indian shoppers actually type, not clean English

A D2C retail brand

Deployed Vespa, the engine behind Perplexity and Spotify, to rank and recommend in one pass, results ordered by intent and behaviour.

We Build Search That Sells, Not Just Finds

Relevant results for each customer, ranked on intent, behaviour and language, with recommendations woven in

Vernacular & Voice Search

Search that understands Hinglish, code-mixing, voice and regional languages, so customers find what they want faster.

The world's largest sporting goods retailer

Built and tuned Algolia search with vernacular synonyms and typo-tolerance, matched to how Indian shoppers actually type, not clean English

Intent & Behaviour Ranking

Intent-led ranking that blends search, behaviour and recommendations to surface the right products first.

A D2C retail brand

Deployed Vespa, the engine behind Perplexity and Spotify, to rank and recommend in one pass, results ordered by intent and behaviour.

Merchandised Recommendations

Customers find more than they searched for — your recommendation logic, rendered in the results, with rules you control.

"Most on-site search still ranks for the catalogue, not the customer, and in India it doesn't even understand how people type: Hinglish, code-mixed, half-spoken. We build search that reads the language and ranks on intent, then renders your merchandising logic right in the results.

We did it on Vespa, Algolia and custom NLU, for a D2C retailer and the world's largest sporting goods retailer. Search stops filtering and starts selling"

"Most on-site search still ranks for the catalogue, not the customer, and in India it doesn't even understand how people type: Hinglish, code-mixed, half-spoken.

We build search that reads the language and ranks on intent, then renders your merchandising logic right in the results.

We did it on Vespa, Algolia and custom NLU, for a D2C retailer and the world's largest sporting goods retailer. Search stops filtering and starts selling"

- — Avinash Reddy, Founder & CEO, ONE CX™

Built on the Search Platform That Fits Search Platform That Fits

We build search on the platform that fits your scale and stack, managed, open-source or custom, with the India-language layer on top.

Our Expertise

Vespa Search

Our pick when search must rank and recommend in one engine, the platform behind Perplexity and Spotify, built for relevance at scale. We built on it for Voyze, a thought-leadership content engine.

Algolia

Our pick for fast, managed search, zero-ops typo-tolerance and ranking, tuned with vernacular synonyms. We built and tuned it for the world's largest sporting goods retailer, matched to how Indian shoppers actually type.

Elasticsearch / OpenSearch

Our pick for scale and control, full ownership of retrieval and relevance, on your infrastructure. Deployed for a leading talent marketplace.

Vertex AI - Search for commerce

Our pick on Google Cloud, managed semantic retail search, with the India-language layer tuned on top.

Typesense / Meilisearch

Our pick for lightweight, open-source search, fast to ship, easy to own.

Custom NLU Layer

Our pick when vernacular is the edge, code-mixing and intent models built on your data, owned by you.

Baseline Architecture

Query In

A customer searches, typed or spoken, in English, Hinglish or a regional language, from web or app.

Language & Intent

We read the language and the meaning code-mixing resolved, 'sasta' understood as cheapest.

Retrieve & Rank

The catalogue is searched on meaning and ranked by intent and behaviour, the right results first.

Merchandise & Recommend

Recommendations woven in, 'bought together', 'you may also like', rendered from your decisioning logic.

Rules & Personalisation

Your merchandising rules and each customer's history shape the final order automated, controllable.

Learn from Every Search

Clicks, taps and zero-results feed back the search sharpens with every query.

Our Expertise

Vespa Search

Our pick when search must rank and recommend in one engine, the platform behind Perplexity and Spotify, built for relevance at scale. We built on it for Voyze, a thought-leadership content engine.

Algolia

Elasticsearch / OpenSearch

Vertex AI - Search for commerce

Typesense / Meilisearch

Custom NLU Layer

Baseline Architecture

Query In

A customer searches, typed or spoken, in English, Hinglish or a regional language, from web or app.

Language & Intent

Retrieve & Rank

Merchandise & Recommend

Rules & Personalisation

Learn from Every Search

Why India's Product Leaders Trust ONE CX With Search

01

We Speak How India Searches

We tune search for Hinglish, code-mixing, voice and regional scripts, the way real customers type and speak, not the clean English search was built for

We Speak How India Searches

We tune search for Hinglish, code-mixing, voice and regional scripts, the way real customers type and speak, not the clean English search was built for

02

Search That Merchandises, Not Just Matches

Algolia-style matching is the foundation we build on. We add intent ranking and recommendations inside the results, so search discovers and sells, it doesn't just filter.

Search That Merchandises, Not Just Matches

Algolia-style matching is the foundation we build on. We add intent ranking and recommendations inside the results, so search discovers and sells, it doesn't just filter.

03

Rules Stay Yours

You control the merchandising, what to boost, bury or bundle, and we automate it. The engine learns; you keep the levers.

Rules Stay Yours

You control the merchandising, what to boost, bury or bundle, and we automate it. The engine learns; you keep the levers.

04

This Is On-Site Search, Not SEO

We personalise the search inside your product, not how you rank on Google. Your customers are already here; we make sure they find, and buy.

This Is On-Site Search, Not SEO

We personalise the search inside your product, not how you rank on Google. Your customers are already here; we make sure they find, and buy.

What Indian Growth Leaders Ask About Search Personalisation

What does ONE CX actually do in a search personalisation engagement?

We build and personalise the search inside your own website or app, so when a customer searches, the results understand their language and intent, and rank what they want first. Vernacular, voice, code-mixing and personalisation, on your stack, learning from every query.

Isn't this just SEO?

No. SEO is how you get found on Google, off-site discovery. This is the search box inside your product, where customers who are already here look for what to buy. We make those results rank for the person and sell, not just match the keyword. Different surface, different goal.

What happens to searches that return nothing today?

A zero-result search is a customer who wanted to buy and couldn't find it, a lost sale you never see. ONE CX closes the gaps with synonyms, spell-tolerance and vernacular understanding, so far fewer searches dead-end, and more convert.

How is this different from the recommendation engine?

The recommendation engine decides which products are worth showing a customer, anywhere. Search personalisation applies that decision to a query: it reads what the customer typed, ranks the catalogue for them, and renders the engine's recommendations inside the results. The engine decides; search ranks and shows. They share logic, not a job.

Does it work for voice search?

Yes. Voice is how much of India searches now, especially in Tier-2 and Tier-3 markets. We handle spoken queries, conversational phrasing and the accents and code-mixing that come with them, so a voice search finds the same right result a typed one would.

How does the DPDP Act affect search personalisation?

Search personalisation uses a customer's language, history and behaviour, so it runs on personal data, within verifiable, purpose-linked consent, under an Act that carries penalties of up to ₹250 crore for a data breach (DPDP Act, 2023). Only consented signals shape results, with audit trails on what ranked and why. Defensible by design.

How is this different from experience personalisation?

Experience personalisation is the whole content system, what each customer sees across the app. Search is one high-intent surface of it: the results, recommendations and ranking when they look for something. We build both, fenced cleanly.